Ensemble Circuit Configuration Questions

[Sean Costello]

mbartkow at ET.PUT.Poznan.PL wrote:

> > The 2 LFOs with 120 degrees phase shift could be based around a single CD4069
> > IC, as in the Polysix ensemble effect (thanks to Mike I. for telling me about
> > this).
> 
> While direct generating of three sine waves with 120 phase shift maight be tricky
> I personally (as an analogue purist) would avoid digital circuits as much as possi-
> ble. 

The CD4069 is being used in an analog manner - the buffer stages (or
whatever the hell they are) are used as low-cost inverting amplifiers,
or, uh, something. Huh huh. Wow, my brain hurts this morning. Anyway,
the CD4069 solution seems like a very simple way of getting the proper
signals. 

> Passing a white noise through a very narrow band BPF results in a "sine" wave of
> fluctuating amplitude. This perfectly mimics only one aspect of the "natural"
> vibrato, namely the depth instability. There is also a speed instability which
> would be achieved by tuning the center frequency of the filter using another
> 1/f noise. I suppose that similar effect may be achieved by modulating a speed
> and the output amplitude of a VCLFO using the same noise. The latter solution is
> probably cheaper. The depth of both modulations might be manually adjustable.

One caution: Randomness does not necessarily sound "natural." In
particular, using random LFOs may not produce a sound that is in any way
"human." Since you are trying to simulate multiple instruments with
vibrato, this is important.
 
> What about more delay paths ? Is three a magic number or just the cost compromise ?
> I understand that three varied delays is enough mess for the ear to prevent it from
> recognising any modulation regularity when periodic modulation is used. What about
> 4 paths with 90 degrees phase shift or 5 paths with 72 degrees ? OTOH, maybe two is
> okay for random modulation ?

Three is probably the minimum number of delay lines for obtaining this
effect. Four or more lines would work great, although it would be worth
experimenting to see what modulation signals would be useful. I doubt
that 2 delay lines would obtain the effect. No real good reasons, just a
gut feeling. Think about a 2-VCO synth, versus a 3-VCO synth. With 2
VCOs, you will only have one beat frequency pattern, as there are just 2
sources to beat against each other. With 3 VCOs, you have 3 direct beat
frequency patterns (VCO 1 against VCO2, VCO 2 against VCO3, VCO1 against
VCO3), as well as the complex interaction of the beat frequency patterns
with each other. Again, more of a gut feeling/intuitive understanding,
but 3 delay lines seems like a minimum.

> PS. Does anybody know what exactly are the lenghts of the delay lines used in
> vintage staff (especially Solina) ? MN3007 (1024 BBD stages) is acceptably cheap,
> but MN3005 (4096 stages) seem to be unreasonably expensive (8 times the price
> of 3007).

1024 stages is plenty long. Other synths use BBDs with 512 or less
stages (I think the Solina used BBDs with 256 stages, but don't quote me
on that).

Sean Costello